The use of MBTI in Software Engineering
Rien Sach
Maths and Computing
Faculty
Open University
Milton Keynes, MK7 6AA
United Kingdom
[email protected]
Marian Petre
Maths and Computing
Faculty
Open University
Milton Keynes, MK7 6AA
United Kingdom
[email protected]
Helen Sharp
Maths and Computing
Faculty
Open University
Milton Keynes, MK7 6AA
United Kingdom
[email protected]
Keywords: POP-I.A. Group Dynamics, POP-II.A. Personality, POP-VI.F. Exploratory
Abstract
In this paper we evaluate the use of Carl Jung’s theories of Psychological Type assessed using the
Myer-Briggs Type Indicator in the Software Engineering field. The current level of implementation
and its quality is established and the results discussed to provide insight into what we currently know,
and suggestions on what could be important to investigate for the future.
Upon gathering MBTI data from a range of sources it is apparent that there is agreement on the types
of personalities often discovered inside software engineering. Thinking and judging personality
preferences are commonly found, while feeling and perceiving is far less common. This differs
substantially from results representative of the American population, and supports the belief that
software engineers are more commonly represented by specific types of people.
However, there is discrepancy between four of the 16 types identified in the MBTI, suggesting that
there is still some understanding to be gained about personality in software engineering, and we do
not by any means know the exact breakdown of types present within the industry.
1. INTRODUCTION
Software Development has been an expanding market for over 40 years, and it is estimated that the
global software market grew by 6.5% in 2008 and is now valued at $303.8 billion [1]. It is also
predicted that by 2013 the global software market will be valued at $457 billion [1].
Personality is a term used to describe the behaviour, traits and character of an individual, and can be
used to suggest how different individuals process situations and events [5]. Each individual’s different
personality relates to how they prefer to use their mind, and this can explain apparently random
behaviour and differences [3].
The Myers-Briggs Type Indicator has been used for over 50 years to identify the personality type of
an individual and their personality preference, making the theories of Jung useful and applicable to
everyday life.
In this paper we present the compiled results of 5 MBTI assessments on software engineering
practitioners, and what this tells us about the personality of a software engineer. We then proceed to
compare the compiled results with previously published results to draw conclusions and comparisons.
The purpose of this paper is to identify the current level of published data on MBTI assessments
administered specifically to software engineers, their quality, validity, and what they tell us and
suggest about the personalities of software engineering practitioners.
1.1 Background
The Myers-Briggs type indicator was developed by Katherine Briggs and her daughter Isabel Myers
from the theories first published in 1921 by Carl Jung [6]. The MBTI (Myers-Briggs Type Indicator)
was first published in 1962 [7], and has become a widely accessible and used tool in assessing a
person’s personality type.
The Myers-Briggs type indicator has become the most widely used personality inventory, with over
3.5 million assessments administered worldwide each year [7, 8].
ISTJ
ISFJ
INFJ
INTJ
ISTP
ISFP
INFP
INTP
ESTP
ESFP
ENFP
ENTP
ESTJ
ESFJ
ENFJ
ENTJ
Table 1. Possible MBTI Types
Carl Jung’s theory on psychological type states that there are two worlds in which we can focus our
minds (Extraversion, Introversion), when we are using our mind we are either taking in information
(Perceiving) or processing this information and drawing conclusions from it (Judging).
Additionally we can perceive in two ways, by living in the present and focusing on what is real and
actual (Sensing), or looking towards the future and the possibilities (iNtuition). And when we are
judging this can also be done in two ways, by looking at the logical consequences and being analytical
(Thinking), or by looking at what is important to ourselves and others and assessing the impact on
people (Feeling).
These four pairs of scales produce the 16 possible Myers-Briggs types indicated in Table 1, and
Figure 1 represents the four dichotomies and the two different preferences for each one.
According to the theory everyone has a preference to one of each of the paired scales [7], and this leads
to your type category. For example a personality of INFP is someone with preferences to Introversion,
Intuition, Feeling and Perceiving. Additionally, everybody has a favourite process which is used
primarily in their favourite world.
Figure 1. MBTI Dichotomies
Your favourite process is one of the two middle letters, and is most often used in your favourite
world. To balance this, your second favourite process is used most often in the other world. For
example, an INTJ would be described as favouring introverted intuition and extroverted thinking.
A lot of significance is put into your personality type as it suggests how you process and gather
information, how you may act in situations, and your preferences in career choice [3]. It has been
reported that the personality of a software engineer and the entire team is an important factor relating
to project success and team cohesion [4].
There has been some widely publicised criticism of the MBTI assessment, stating that:
There was no support for the view that the MBTI measures truly dichotomous preferences or
qualitatively distinct types, instead, the instrument measures four relatively independent
dimensions.
(R. McCrae and P. Costa, 1989)
This claim and other claims regarding the MBTI assessment, such as a lack of independent evidence,
and no evidence that MBTI measures truly dichotomous preferences or qualitatively distinct types,
among other criticisms, were published in 1989 [22].
Additionally, reported MBTI results in software engineering have not been without their problems. S.
McDonald and H. M Edwards (2007) reported identifying an article which had claimed to be
reporting MBTI assessment results was not actually using a MBTI assessment [10], as later admitted by
the initial authors in the technical report [21].
1.2 Literature Review
The literature review consisted of searching through the major online databases of journals and papers
(including IEEE, ACM, PsycINFO) specifically searching for MBTI, Myers-Briggs personality, and
software engineering.
The results displayed a vast amount of papers discussing personality and its usage [10] and potential
effect [9, 11, 12] on a range of aspects of software engineering, such as educating engineers [13], and
practitioner preferences [14]. However there is a significant lack of published complete MBTI data
specific to software engineers, as many of the published works only print their conclusions and what
the data informs them of, and not the actual MBTI preference breakdowns.
Some of the papers focused on alternative approaches to identifying and describing personality, such
as the Five Factor Model [19, 15], as well as some other papers focusing on specific practices or roles in
software engineering such as pair programming [15, 16, 18], software testing [17], and software team
cohesion [19].
The literature review identified 12 papers reporting tables of Myers-Briggs Type Indicator data. The
following section discusses these papers.
1.3 Collected MBTI Data
Of the 12 identified published data collections, there were 10 useable collections. One collection of
data was removed as it was not possible to access enough details about how the data was collected
and when it was collected. Of the remaining 10 collections of data there is 5 that collect their data
primarily from practitioners, and 5 that collect their data primarily from students.
It was decided to focus on the papers that specifically collected their data from active working
practitioners and to exclude data collected from predominantly student samples. This was to ensure
that the data was representative of working software practitioners, and not people who were only
potential software engineers. The paper will now continue to discuss these 5 data samples.
The 5 pieces of data are predominantly published in a span of 5 years, with the first 4 all ranging from
1985 to 1990 and the 6th data being published in 2003. All of them were collected from western
companies, primarily from America. We’ll now look at the information provided by each source
separately in chronological order.
The first source is an article published in a computing magazine (Datamation) in 1985 by M.L. Lyons.
The data consisted of 1,229 computer professionals, of whom 213 of them were based in the UK and
Australia; the other 1,016 were based in America. The data consisted of 73% males and 27% females,
and the median age for males was 34 and for females it was 31.
The surveyed members had a median of seven years experience in computing, with 30% having
worked in computing for 5 years or less. Twenty percent of the surveyed members were in
management positions, with an additional 20% working as project managers and team leaders. No
clear information is given about the gathering of the data.
The second source was an article published in the Journal of Psychological Type in 1988 by E. A.
Buie. In this study the MBTI results of 47 scientific computer professionals were presented, with
57.6% of them being male. The MBTI method used to gather the data from this source is described as
being from a questionnaire “developed specifically for this study”.
The third source was published in the Journal of Psychological Type in 1988 by P. Westbrook. The
results presented were from a group of 153 professionals described as “Field Engineers”. The results
were said to be gathered from a “Fortune 400 computer company”, and describes the method of
gathering the information as a “self-scoring short form”.
The fourth source is an article published in the ACM SIGCPR journal in 1989 by D. C. Smith. The
data presented was gathered from 37 systems analysts working at a large insurance company. The
method described for gathering the results is the “shorter version of the MBTI” and also states the
questionnaire was administered by a psychologist.
The fifth source is an article published in the International Journal of Human-Computer Studies in
2003 by L. F. Capretz. This data collection contained 100 software engineers of which 80% were
male and 20% female, and states that they were either working for the government, working for
software companies, or were studying in public or private universities. The published paper also states
that MBTI assessment was administered using Form G, which is an older and less reliable form of
administering the MBTI assessment.
2. FINDINGS
This section discusses the 5 papers implementing a MBTI assessment on a group of software
engineering practitioners. The data will be presented in this section, and discussed in subsequent
sections.
Figure 2. Total MBTI Results
2.1 Compiled Results
The purpose of these studies was to identify the MBTI preferences of the participants, and these
combined results from the 5 collections of data are expressed as percentages of the 16 different types.
The results gathered from the 5 sources of MBTI data are displayed in figure 2. It is apparent that
widespread conformity can be noted on the majority of types, with the most variation in reported
levels being represented by ISTJ, INTJ, INTP, and ESTJ.
ISTJ
ISFJ
INFJ
INTJ
6.47
2.29
3.56
3.97
ISTP
ISFP
INFP
INTP
2.33
2.28
2.56
4.41
ESTP
ESFP
ENFP
ENTP
2.48
2.62
1.97
2.85
ESTJ
ESFJ
ENFJ
ENTJ
10.55
1.57
0.95
2.69
Table 2. MBTI Standard Deviation
Table 2 further supports the level of agreement in commonly reported results by presenting the
standard deviation value for each of the 16 MBTI types when all of the different data collection
results are combined.
Figure 3 compared the averages of all of the MBTI types based on the results gathered from the 5
sources to the average preferences of the US population as reported by CAPT (Centre for
Applications of Psychological Type) [2].
The comparison makes it clear that the common preferences of the US population are not reflected
inside software engineering. In different categories the preferences are over or under represented, but
there is an emerging pattern that the thinking preference is consistently over-represented in the
reported MBTI results.
ISTJ, INTJ, ISTP, INTP, ENTP, ESTJ, and ENTJ all display an average higher than the general US
population, with the only exception being the marginal difference on ESTP from 4.30 to 3.76.
Table 3 shows the representation of each pair in the results, as well as the level of standard deviation
and the mean. These figures were generated from the data represented in Figure 2. There is a higher
level of variation when the data is constructed like this and thus the means are less reliable.
Figure 3. MBTI Results and US Population Averages Comparison
As you can see from Table 3, five of the highlighted pairings are pairings including thinking, and four
of the pairings including judging. Sensing is also present four times, and extroversion and introversion
appear three times each. However, feeling does not appear once, neither does perceiving, showing a
clear lack of a common representation to either of these preferences.
Data 1
Data 2
Data 3
Data 4
Data 5
Standard Deviation
Mean
TJ
55.8
42.7
62.1
75.6
50
12.52
57.24
ST
39.2
36.2
56.9
70.2
55
13.93
51.5
IT
55.4
55.4
36.6
51.3
47
7.82
49.14
SJ
36.8
29.9
53
75.6
45
17.67
48.06
IJ
44.7
44.8
29.4
51.3
34
8.91
40.84
IS
33.2
32
32.7
45.9
39
5.92
36.56
ET
25.4
12.8
45.8
37.8
34
12.62
31.16
NT
41.6
32
25.5
18.9
26
8.53
28.8
EJ
21.1
12.8
39.9
35.1
24
10.92
26.58
ES
13.1
14.9
34.7
35.1
28
10.59
25.16
IN
33.9
42.6
11.8
13.5
18
13.59
23.96
TP
25
25.5
20.3
13.5
31
6.55
23.06
IP
22.4
29.8
15.1
8.1
23
8.31
19.68
NJ
29
27.7
16.3
10.8
13
8.45
19.36
NP
24.7
25.6
16.4
8.1
20
7.12
18.96
EN
19.8
10.7
20.9
5.4
15
6.45
14.36
SP
9.5
17
14.4
5.4
22
6.46
13.66
EP
11.8
12.8
15.7
5.4
19
5.06
12.94
IF
11.7
19.2
7.9
8.1
10
4.64
11.38
SF
7.1
10.7
10.5
10.8
12
1.84
10.22
FJ
10
14.9
7.2
10.8
8
3.01
10.18
FP
9.2
17.1
10.5
0
11
6.15
9.56
NF
12.1
21.3
7.2
0
7
7.87
9.52
EF
7.5
12.8
9.8
2.7
9
3.71
8.36
Table 3. MBTI Paired Preferences
The level of representation of thinkers is very apparent here with multiple pairings including thinking
surpassing an average representation of 50%.
3. RESULTS SUMMARY
For the most part the MBTI results show agreement with each other. There are only two different
types with a standard deviation above 5.00 (ESTJ with 6.47 and ISTJ with 10.55).
Thinkers are present significantly more within the results than feelers. This can be seen within the
results where the types of feelers are commonly scoring lower than the types of thinkers.
4. DISCUSSION
4.1 Results Discussion
The results presented here, although from a range of sources and publications, present a common view
on the MBTI personalities present inside software engineering. Although four of the five collections
of data were during 1985-1990, it is interesting that the one collection of data from 2003 does not
show much if any of a change in the reported types.
The data suggests a relationship between psychological type and software engineers, but this does not
imply that there is a singular fix for a software engineer nor that one type is more useful than any
other, it simply suggests that there are some types more prominently found inside software
engineering.
Specifically thinkers have been established by the data as being commonly represented as a
preference by the majority of the reported results, with preferential thinkers representing an average
(mean) of 80.3% of the reported results, compared to 19.7% being feelers.
There is also a 60/40 split in favour of introverted preference than the extroverted preference, which is
different to the near 50/50 split suggested to be present among the US population. There is also a 67%
representation of the judgement preference being present over the perceiving preference.
It is also worth noting that the pairing of thinking and judging is substantially higher than the US
population (24.20%) with an average of 57.24%, while the pairing of feeling and perceiving is much
lower than the US population (29.80%) with an average reported result of 9.56%.
All of this information leads to suggesting that thinkers and judgers are more attracted to software
engineering and feelers and perceivers are less attracted to software engineering jobs, based on the
data combined from 5 sources of MBTI assessment.
4.1 Limitations
The information presented here has inherent limitations on how the original collections of data have
been published. There is not enough information published in most of the papers detailing how the
assessments were administered, if a psychologist was present to administer the MBTI assessments, or
what form was used. This makes it impossible to ensure the data is comparable across studies.
The details of the respondents also vary, with some of the papers stating the exact breakdown of age,
gender and experience, with others offering either incomplete breakdowns or no information at all.
The physical number of respondents also ranges from over a thousand to under fifty, as well as the
type of people the results report about.
It would be acceptable to classify the 2nd through to the 5th source as valid as software engineering
personnel, but the first source is simply too generic to be classified as specifically about software
engineers. This means the type of data being compared cannot be described as being exclusively
focused on identical groups of software engineers in a range of studies.
5. CONCLUSIONS
The results gathered here represent a common picture on the majority of the MBTI preferences
present inside software engineering, and show a large preference towards thinkers and judgers.
On average there are 57.24% of the respondents with a thinking judging preference, 51.50% with a
sensing thinking preference, and 49.14% showing an introverted thinking preference, further
establishing the strong preference towards thinking in the results presented here.
The combined results of these 5 MBTI data collections compare well with other publications where
full MBTI data was not available but the conclusions were presented. Bush and Schkade (1985) [23]
also identified thinkers as being a significantly represented preference in their results of 40
programmer analysts with 73% of the respondents reported as being thinkers. They also identified that
70% of their respondents preferred judging as opposed to perceiving.
Thomsett (1990) [24] identified a high representation of thinkers (79%) in a group of 656 computer
professionals. It was identified that there was a far higher representation of judgers than there was
perceivers, with 92.3% of the respondents showing a preference towards judging.
However it is clear that there has been very little work done on the types of personalities in software
engineering practitioners recently using the MBTI assessment tool, with only one of the results being
published in the past 20 years (2003). This is important, as it’s very likely at the types of roles, and
software engineering has changed in those 20 years, along with the years since that study, with the
growth of agile development practices.
If we are to accept and believe what we are told about the implications of personality and MBTI, it is
extremely important that further work is done to understand the types inside software engineering. A
better understanding could lead to a better workplace and task fit to specific people, specifically a
greater understanding into the effects of personality on teams. It is not unreasonable to consider the
possibility that the dynamics of a team and the personalities present could drastically affect
performance and productivity.
The results presented here are old and generic, one collection of data is even too generic to really be
considered about software engineers. Future work on personality, and specifically MBTI, should
focus on establishing up-to-date classifications of what software engineering really means, and
specifically the different roles identifiable inside this area. It’s quite possible that the role of
somebody considered a software engineer could also affect the type of personality drawn to this role,
and potentially explain the variations reported in this paper. We do not yet have any data to suggest
that the skills and personality required to be a software designer are equal to that of the skills required
to be a software programmer, and Capretz (2010) [25] recently reports on the potential varying
personality types required under the umbrella of software engineering.
6. REFERENCES
1. Datamonitor (2006) Software: Global Industry Guide. Available from:
http://www.infoedge.com/product_type.asp?product=DO-4959, accessed 01/02/2010.
2. CAPT (Centre for Applications of Psychological Type) (2010). Estimated Frequencies of the
Types in the United States Population. Available from: http://www.capt.org/mbtiassessment/estimated-frequencies.htm, accessed 25/01/2010.
3. Myers, I.B. 1998. Introduction to type : a guide to understanding your results on the MyersBriggs Type Indicator.5th Edition. Oxford : Oxford Psychologists Press, 1998.
4. J. Karn and T. Cowling, “A follow up study of the effect of personality on the performance of
software engineering teams,” in Proceedings of the 2006 ACM/IEEE international symposium
on Empirical software engineering, 2006, 241.
5. Sidney J. Blatt, “Where Have We Been and Where Are We Going? Reflections on 50 Years
of Personality Assessment,” Journal of Personality Assessment 50, no. 3 (1986): 343.
6. C.G. Jung, Psychological Types, 1st ed. (Routledge, 1992).
7. Isabel Briggs Myers, An Introduction to Type: A Guide to Understanding Your Results on the
Myers-Briggs Type Indicator: European English Version, 5th ed. (Oxford Psychologists
Press, 198).
8. Catherine Bishop-Clark and Daniel D. Wheeler, “The Myers-Briggs personality type and its
relationship to computer programming.,” Journal of Research on Computing in Education 26,
no. 3 (Spring94 1994): 358.
9. Alessandra Devito Da Cunha and David Greathead, “DOES PERSONALITY MATTER? AN
ANALYSIS OF CODE-REVIEW ABILITY.,” Communications of the ACM 50, no. 5 (May
2007): 109-112.
10. S. McDonald and H. M Edwards, “Who should test whom?,” Communications of the ACM
50, no. 1 (2007): 71.
11. Robert Feldt et al., “Links between the personalities, views and attitudes of software
engineers,” Information and Software Technology In Press, Accepted Manuscript.
12. J. S. Karn et al., “A study into the effects of personality type and methodology on cohesion in
software engineering teams.,” Behaviour & Information Technology 26, no. 2 (March 2007):
99-111.
13. Lucas Layman, Travis Cornwell, and Laurie Williams, “Personality types, learning styles, and
an agile approach to software engineering education,” ACM SIGCSE Bulletin 38, no. 1 (3,
2006): 428.
14. L Capretz, “Personality types in software engineering,” International Journal of HumanComputer Studies 58, no. 2 (2, 2003): 207-214.
15. N. Salleh et al., “An empirical study of the effects of personality in pair programming using
the five-factor model,” in Proceedings of the 2009 3rd International Symposium on Empirical
Software Engineering and Measurement, 2009, 214–225.
16. J. E Hannay et al., “Effects of Personality on Pair Programming.”
17. Shoaib, L.; Nadeem, A.; Akbar, A.; , "An empirical evaluation of the influence of human
personality on exploratory software testing," Multitopic Conference, 2009. INMIC 2009.
IEEE 13th International , vol., no., pp.1-6, 14-15 Dec. 2009
18. T. Walle and J. E Hannay, “Personality and the nature of collaboration in pair programming,”
in Proceedings of the 2009 3rd International Symposium on Empirical Software Engineering
and Measurement, 2009, 203–213.
19. Rebecca H. Rutherfoord, “Using personality inventories to help form teams for software
engineering class projects,” ACM SIGCSE Bulletin 33, no. 3 (9, 2001): 73-76.
20. R. R McCrae and O. P John, “An introduction to the five-factor model and its applications,”
Personality: critical concepts in psychology 60 (1998): 295.
21. Karn, J.S. and Cowling, A.J. A study into the effect of disruptions on the performance of
software engineering teams. Research Memorandum CS-04-07, Department of Computer
Science, Sheffield University, 2004.
22. Robert R. McCrae and Paul T. Costa Jr., “Reinterpreting the Myers-Briggs Type Indicator
From the Perspective of the Five-Factor Model of Personality.,” Journal of Personality 57,
no. 1 (March 1989): 17-40.
23. Chandler M Bush and Lawrence L Schkade, “In search of the perfect programmer,”
Datamation 31, no. 6 (1985): 128-132.
24. R. Thomsett, "Building Effective Project Teams", American Programmer, Summer 1990
25. Luiz Fernando Capretz and Faheem Ahmed, “Making Sense of Software Development and
Personality Types,” IT Professional, 2010.
7. MBTI DATA SOURCES
1. Michael L. Lyons, “THE DP PSYCHE.,” Datamation 31, no. 16 (1985): 103-105, 108, 110.
2. Buie, E. A. (1988). Psychological type and job satisfaction in scientific computer
professionals. Journal of Psychological Type, 15, 50-53.
3. Westbrook, P. (1988) Frequencies of MBTI Types Among Computer Technicians. Journal of
Psychological Type, 15, 49.
4. D. C. Smith, “The personality of the systems analyst: an investigation,” SIGCPR Comput.
Pers. 12, no. 2 (1989): 12-14.
5. L Capretz, “Personality types in software engineering,” International Journal of HumanComputer Studies 58, no. 2 (2, 2003): 207-214.
6. R. Thomsett, "Building Effective Project Teams", American Programmer, Summer 1990.
7. Jo Ann C. Carland and James W. Carland, “Cognitive styles and the education of computer
information systems students.,” Journal of Research on Computing in Education 23, no. 1
(Fall90 1990): 114.
8. Eugene Kaluzniacky, Managing psychological factors in information systems work (Idea
Group Inc (IGI), 2004).
9. L. F Capretz, “Clues on Software Engineers' Learning Styles,” International Journal of
Computing & Information Sciences 4, no. 1 (2006): 46-49.
10. J. S. Karn et al., “A study into the effects of personality type and methodology on cohesion in
software engineering teams,” Behaviour & Information Technology 26, no. 2 (3, 2007): 99111.
11. S. Holmes and P. L Kerr, (2007). The IT crowd. Australian Pyschological Type Review 9, no.
1 (4, 2007): 31-38.